The hardest sudokus (new thread)

Everything about Sudoku that doesn't fit in one of the other sections

Re: The hardest sudokus (new thread)

Postby dxSudoku » Mon Aug 30, 2021 12:47 am

I have another question if someone is willing to answer it. If you have 22 givens in your constellations of givens, and given that there are 81 cells, how many unique patterns of constellations are there with 81 givens. Is this simply 81 choose 2 combinatorics? 2 meaning either the given is present in the constellation of givens or the cells is empty. I'm just okay at math so I'm not that confident in my thinking.
dxSudoku
 
Posts: 43
Joined: 06 April 2020

Re: The hardest sudokus (new thread)

Postby marek stefanik » Mon Aug 30, 2021 7:43 am

There is no (known) sudoku that would require brute-force as a solving step – otherwise you couldn't get a meaningful rating function. Most rating systems add another layer of chains to solve the hardest puzzles.
Only puzzles with exactly one solution are considered valid – this is checked for before attempting to rate the puzzle and it is indeed done by brute-force.
For more insight into the creation process, check out this post by mith.

dxSudoku wrote:If you have 22 givens in your constellations of givens, and given that there are 81 cells (...) either the given is present in the constellation of givens or the cells is empty.
You have 81 choose 22 ways to organise them, some of which cannot produce a unique solution when filled with digits (such as all givens in one band).
marek stefanik
 
Posts: 359
Joined: 05 May 2021

Re: The hardest sudokus (new thread)

Postby mith » Wed Sep 01, 2021 2:48 pm

Code: Select all
31c
........1.....2.3..34..5.....1.6..736.34.781.78....6....8...1.631..86.4.4.67...8.  ED=11.6/1.2/1.2


This one isn't too surprising since we were already in +dfc range, but finally found one.
mith
 
Posts: 996
Joined: 14 July 2020

Re: The hardest sudokus (new thread)

Postby mith » Wed Sep 01, 2021 2:52 pm

Code: Select all
36c
...........1..2.34..2.1356..17.86..3.6.2.1.8.2.873.6...7..28...12.36..7.8.61.73..  ED=10.6/1.2/1.2


(A few 35c as well.)
mith
 
Posts: 996
Joined: 14 July 2020

Re: The hardest sudokus (new thread)

Postby urhegyi » Fri Sep 03, 2021 2:43 pm

What is the highest sudoku rating ever? I thought it was below the theoretical 12.0. But now i'm confused. I just rated a sudoku-x (with diagonal blocks) for over 18,5 hours and this was the result:
Hidden Text: Show
Image

re-rated in 1.5 hours with the newest version:
Hidden Text: Show
Image
Last edited by urhegyi on Mon Sep 13, 2021 7:05 pm, edited 1 time in total.
urhegyi
 
Posts: 748
Joined: 13 April 2020

Re: The hardest sudokus (new thread)

Postby mith » Fri Sep 03, 2021 3:10 pm

11.9 is the high for classic. With additional constraints, you can go higher (anti-knight is another with a puzzle known to be at least 12.6 but probably higher).
mith
 
Posts: 996
Joined: 14 July 2020

Re: The hardest sudokus (new thread)

Postby 999_Springs » Fri Sep 03, 2021 4:15 pm

eleven wrote:

It's really impressing, that about 10 years after people in this forum posted free programs, which solve each known puzzle (100 times harder than the author's ones) in a split second, a paper is written where it says
"No puzzle requiring more than 3 minutes (in a relatively old computer) was found."

just saw this. this is hilarious. i checked it out and solved their "most difficult puzzle" by hand in 12.5 minutes using only hidden pairs and locked candidates. a speed solver on here like strmckr or qiuyanzhe or maybe sclt might be able to beat the computer's best time - and by the standards of that paper that would mean humans are better than computers at sudoku :lol:

heetbeet seems to also be interested in some sudoku variants like killer sudokus:
heetbeet wrote:We also used a ton of Killer Sudokus, Calcudokus, Kakuros and Fill-a-Pix for further colouring in our puzzle database.

i would also be interested in what happens with those; there is no widely accepted means of defining difficulty of a killer sudoku other than time to brute force, or number of guesses needed. unlike normal sudokus which solve with brute force in microseconds, the puzzles in this thread need a few minutes to solve with brute force :o and unlike this author's example these killer puzzles seem to be impossible for humans too
999_Springs
 
Posts: 591
Joined: 27 January 2007
Location: In the toilet, flushing down springs, one by one.

Re: The hardest sudokus (new thread)

Postby RSW » Fri Sep 03, 2021 8:30 pm

999_Springs wrote:just saw this. this is hilarious. i checked it out and solved their "most difficult puzzle" by hand in 12.5 minutes using only hidden pairs and locked candidates. a speed solver on here like strmckr or qiuyanzhe or maybe sclt might be able to beat the computer's best time - and by the standards of that paper that would mean humans are better than computers at sudoku :lol:

Just tried it on my solver, which I'm sure is nowhere near as efficient as the solvers written by other forum members. My solver took 110 milliseconds using basics only.
Solution:
Hidden Text: Show
Singles: 1r6c9 -1r123c9 1r7c6 -1r9c46 -1r123c6 1r9c3 -1r123c3 2r7c5 -2r8c46 -2r12356c5 2r5c9 -2r5c46 -2r123c9 4r8c6 -4r123c6 -4r8c89 5r4c8 -5r4c1346 -5r789c8 1r1c8 -1r23c8 2r6c6 -2r123c6 2r1c3 -2r23c13 -2r8c3 2r8c1

Subset: Column 4, (579) r4c4 r8c4 r9c4 => -79r2c4 -579r3c4 -579r5c4
Subset: Column 4, (3579) r4c4 r5c4 r8c4 r9c4 => -3r2c4 -3r3c4
Subset: Column 7, (489) r4c7 r5c7 r6c7 => -4r1c7 -489r3c7 -49r7c7
Subset: Column 7, (3489) r4c7 r5c7 r6c7 r7c7 => -3r1c7 -3r3c7
Subset: Block 3, (467) r1c7 r1c9 r3c7 => -4r2c8 -467r2c9 -4r3c8 -467r3c9
Subset: Block 3, (46789) r1c7 r1c9 r2c9 r3c7 r3c9 => -89r2c8 -89r3c8
Subset: Block 4, (35689) r4c1 r5c1 r6c1 r6c2 r6c3 => -69r4c3 -359r5c3

Singles: 3r5c4 -3r5c16 3r7c7 -3r7c28 -3r9c8

Box/Line: Row 8/Box 9, (8)r8c8 r8c9 => -8r9c89
Box/Line: Column 2/Box 1, (4)r1c2 r2c2 r3c2 => -4r23c3
Box/Line: Column 2/Box 1, (7)r1c2 r2c2 r3c2 => -7r23c3
Box/Line: Column 4/Box 8, (5)r8c4 r9c4 => -5r9c56
Box/Line: Column 7/Box 3, (6)r1c7 r3c7 => -6r1c9
Box/Line: Column 7/Box 3, (7)r1c7 r3c7 => -7r1c9
Box/Line: Column 8/Box 9, (4)r7c8 => -4r7c9
Box/Line: Column 8/Box 9, (8)r8c8 => -8r8c9
Box/Line: Column 8/Box 9, (9)r7c8 r8c8 r9c8 => -9r789c9
Box/Line: Box 3/Row 1, (4)r1c9 => -4r1c25
Box/Line: Box 4/Row 6, (9)r6c2 r6c3 => -9r6c57
Box/Line: Box 7/Column 2, (3)r9c2 => -3r1236c2

Singles: 4r1c9 8r6c7 -8r6c125 -8r45c7 5r7c9 -5r7c2 -5r89c9 9r7c2 -9r69c2 -9r8c23 -9r7c8 4r7c8 9r9c8 -9r9c456 -9r8c8 8r8c8

Subset: Row 6, (56)r6c2 r6c5 => -56r6c13
Subset: Row 8, (56)r8c2 r8c3 => -5r8c4 -6r8c9
Subset: Row 9, (78)r9c5 r9c6 => -7r9c49
Subset: Column 2, (56)r6c2 r8c2 => -56r139c2 -6r2c2
Subset: Block 8, (78)r9c5 r9c6 => -7r8c4

Singles: 7r1c2 -7r23c2 -7r1c567 6r1c7 -6r1c56 -6r3c7 7r3c7 -7r3c56 5r1c5 -5r56c5 -5r3c56 -5r1c6 6r6c5 -6r6c2 -6r23c5 -6r4c6 5r6c2 -5r5c1 -5r8c2 6r8c2 -6r8c3 5r8c3 -5r3c3 8r5c1 -8r234c1 -8r5c56 6r4c1 -6r23c1 3r1c6 -3r23c6 3r6c1 -3r23c1 -3r6c3 1r2c1 -1r3c1 -1r2c4 5r3c1 2r2c4 -2r3c4 -2r2c8 1r3c4 3r2c8 -3r2c3 -3r3c8 2r3c8 6r2c3 -6r3c3 -6r2c6 3r3c3 9r6c3 9r8c4 -9r4c4 7r4c4 -7r4c36 -7r5c56 4r4c3 -4r5c3 -4r4c7 7r5c3 9r4c7 -9r4c6 -9r5c7 4r5c7 8r4c6 -8r9c6 7r9c6 -7r9c5 -7r2c6 9r2c6 -9r3c56 -9r2c59 -9r5c6 6r3c6 5r5c6 8r2c9 -8r2c2 -8r3c9 9r3c9 4r2c2 -4r3c2 -4r2c5 8r3c2 7r2c5 4r3c5 8r9c5 9r5c5 7r8c9 3r9c2 5r9c4 6r9c9

Total solution time: 110.86 milliseconds


BTW, I would argue that the term "brute force" has been used rather inaccurately in this discussion. In a puzzle that has been solved using logical moves until an impasse is reached, and then the solver has no remaining strategies except to make a guess, this is not a brute force move. It is a trial and error move, or T&E. To me, brute force refers to a solving technique which never involves any of the recognized Sudoku logic based moves, but rather makes nothing but guesses from the very first move. Brute force is a subset of T&E. Maybe I'm splitting hairs, because experienced human players are generally against any sort of T&E move, brute force or not.
RSW
 
Posts: 669
Joined: 01 December 2018
Location: Western Canada

Re: The hardest sudokus (new thread)

Postby m_b_metcalf » Fri Sep 10, 2021 9:24 am

urhegyi wrote:What is the highest sudoku rating ever? I thought it was below the theoretical 12.0. But now i'm confused. I just rated a sudoku-x (with diagonal blocks) for over 18,5 hours and this was the result:

This puzzle is not new and the result is wrong, see http://forum.enjoysudoku.com/x-sudoku-extreme-t34714-15.html#p309280. Mike
User avatar
m_b_metcalf
2017 Supporter
 
Posts: 13624
Joined: 15 May 2006
Location: Berlin

Re: The hardest sudokus (new thread)

Postby mith » Sat Sep 11, 2021 8:40 pm

Code: Select all
33c
........1....23....24.....5.67..4.8.3.2.86...84..7.6.3.76.38...28.74.3.64.3..2.7.  ED=11.3/1.2/1.2


(Also some new 11.7s in the 22-23c range, and a ton of 11.6s in the 27-30c range. I'll post a more complete update once I get through the current good neighborhood.)
mith
 
Posts: 996
Joined: 14 July 2020

Re: The hardest sudokus (new thread)

Postby 999_Springs » Sun Sep 12, 2021 11:19 am

here is an old-fashioned ascii histogram of chain lengths of the 11.8 and 11.9 puzzles in 1to9only's post (by the way, thanks for doing this!)
Code: Select all
node count rounded down to nearest 16
SE  nodes | sudokus
----------+----------------------
11.8  256 | #########
      272 | ##############
      288 | ################
      304 | #########
      320 | #####
      336 | ###
      352 | ##
      368 | #     Mike's 20
----------+----------------------
11.9  384 | #####
      400 |
      416 | #     Golden Nugget
      432 | #     eleven's Patience
      448 |
      464 | ##    champagne dry; Obi-Wahn 2015
      480 |
      496 |
----------+----------------------
12.0  512 |

from this we can see that:
  • 5 of the 9 11.9's only barely make it into the 11.9 category, with a longest chain of less than 400 nodes, putting them into the lowest 1/8th of possible longest chain lengths for 11.9's (384 is the threshold)
  • no puzzle exists in the top 1/4 of possible longest chain lengths for 11.9's
  • the closest puzzle to SE 12.0 is Obi-Wahn's 11.9, which is the most recently found 11.9, found in 2015
  • 81% of 11.8's are in the lower half of possible longest chain lengths for 11.8's
  • Mike's 20 clue is the 11.8 that is closest to 11.9

mith wrote:
Code: Select all
33c
........1....23....24.....5.67..4.8.3.2.86...84..7.6.3.76.38...28.74.3.64.3..2.7.  ED=11.3/1.2/1.2

this starts with 4 hidden singles so when you fill them in you break Mauricio's 14 year old record from January 2007 for the hardest puzzle with 37 clues, which was his symmetric 11.2. nice!

RSW wrote:BTW, I would argue that the term "brute force" has been used rather inaccurately in this discussion.

i referred to "brute force" only in the context of solving killer sudokus, and called it that because that was what the solvers called it in the thread that i linked. i don't have those killer sudoku solvers installed but my understanding is that what they describe as "brute force" agrees with what you are saying, in other words, nested guesses using DFS or something similar to determine puzzle validity as quickly as possible. in this context, the killer sudokus in that thread take a few minutes to determine that the solution is unique. here is an example of a browser based brute force killer sudoku solver which is the fastest one i know of
999_Springs
 
Posts: 591
Joined: 27 January 2007
Location: In the toilet, flushing down springs, one by one.

Re: The hardest sudokus (new thread)

Postby mith » Sun Sep 12, 2021 4:51 pm

999_Springs wrote:
mith wrote:
Code: Select all
33c
........1....23....24.....5.67..4.8.3.2.86...84..7.6.3.76.38...28.74.3.64.3..2.7.  ED=11.3/1.2/1.2

this starts with 4 hidden singles so when you fill them in you break Mauricio's 14 year old record from January 2007 for the hardest puzzle with 37 clues, which was his symmetric 11.2. nice!


Yeah, I need to run my singles expander on all these hard puzzles, I've probably broken some more of the most-clues-for-given-ER by now.
mith
 
Posts: 996
Joined: 14 July 2020

Re: The hardest sudokus (new thread)

Postby RSW » Sun Sep 12, 2021 4:57 pm

999_Springs wrote:
RSW wrote:BTW, I would argue that the term "brute force" has been used rather inaccurately in this discussion.

i referred to "brute force" only in the context of solving killer sudokus, and called it that because that was what the solvers called it in the thread that i linked. i don't have those killer sudoku solvers installed but my understanding is that what they describe as "brute force" agrees with what you are saying, in other words, nested guesses using DFS or something similar to determine puzzle validity as quickly as possible. in this context, the killer sudokus in that thread take a few minutes to determine that the solution is unique. here is an example of a browser based brute force killer sudoku solver which is the fastest one i know of


Actually, I was referring mainly to dxSudoku's use of the term in both this thread and another recent thread. But, as I said, I'm probably splitting hairs.
RSW
 
Posts: 669
Joined: 01 December 2018
Location: Western Canada

Re: The hardest sudokus (new thread)

Postby hendrik_monard » Mon Sep 13, 2021 3:17 pm

mith wrote:11.9 is the high for classic. With additional constraints, you can go higher (anti-knight is another with a puzzle known to be at least 12.6 but probably higher).


The sudoku in the hidden image of the post by urhegyi » Fri Sep 03, 2021 3:43 pm ,with rating 12.6 is the following:
..1...2....23....44........5.3....6..1......5..6..........78.......9.....7...1.9.
This sudoku has multiple solutions (at least 5000). Isn't it amazing that Sudoku Explainer 1.2.1 tells that "this sudoku can be solved using the following logical methods..."
I would expect that in the case of a sudoku with multiple solutions, no solution can be found with logical reasoning. Or am I missing something?
hendrik_monard
 
Posts: 96
Joined: 19 April 2021
Location: Leuven (Louvain) Belgium

Re: The hardest sudokus (new thread)

Postby m_b_metcalf » Mon Sep 13, 2021 4:31 pm

hendrik_monard wrote:
The sudoku in the hidden image of the post by urhegyi » Fri Sep 03, 2021 3:43 pm ,with rating 12.6 is the following:
..1...2....23....44........5.3....6..1......5..6..........78.......9.....7...1.9.
This sudoku has multiple solutions (at least 5000). Isn't it amazing that Sudoku Explainer 1.2.1 tells that "this sudoku can be solved using the following logical methods..."
I would expect that in the case of a sudoku with multiple solutions, no solution can be found with logical reasoning. Or am I missing something?

Please scroll back to my previous post about this X-sudoku. M.

[Edit: treated as a standard puzzle, it has 3194507 solutions.]
Last edited by m_b_metcalf on Tue Sep 14, 2021 7:02 am, edited 1 time in total.
User avatar
m_b_metcalf
2017 Supporter
 
Posts: 13624
Joined: 15 May 2006
Location: Berlin

PreviousNext

Return to General